Dynamic gesture based short-range human-machine interaction

1. A computer-implemented method for recognizing a dynamic hand gesture, comprising:

detecting an initiation gesture;

starting a gesture recognition engine in response to detecting the initiation gesture;

determining, using the gesture recognition engine, a hand posture in at least one image of a plurality of images;

determining, using the gesture recognition engine, a hand trajectory in the plurality of images;

determining, using the gesture recognition engine, a dynamic hand gesture in response to the hand posture and the hand trajectory;

providing, using the gesture recognition engine, a user interface command in response to determining the dynamic hand gesture

wherein determining the hand trajectory comprises:

determining a plurality of hue-saturation-depth (HSD) histograms; and

tracking, using mean-shift analysis, a moving hand in response to the plurality of HSD histograms.

2. The method of claim 1 , wherein determining the hand posture comprises:

detecting a hand in the at least one image;

segmenting, in response to detecting the hand, the at least one image into a binary image including a hand region; and

determining at least one shape feature in the hand region.

3. The method of claim 2 , further comprising:

determining, using artificial neural multi-layer perceptron (MLP) analysis, a class corresponding to the at least one shape feature.

4. The method of claim 2 , wherein the a least one shape feature comprises at least one of eccentricity, compactness, orientation, rectangularity, width center, height center, minimum box angle, minimum box width, number of defects, difference between left and right portions, or difference between top and bottom portions shape features.

5. The method of claim 2 , wherein detecting the hand comprises using a cascade speeded-up robust feature (SURF) detector to detect the hand.

6. The method of claim 1 , wherein determining the dynamic hand gesture comprises applying a hidden-markov model (HMM) to identify the dynamic hand gesture.

7. A non-transitory computer-readable storage medium comprising a computer program product having stored therein instructions that, if executed, result in:

detecting an initiation gesture;

starting a gesture recognition engine in response to detecting the initiation gesture;

determining, using the gesture recognition engine, a hand posture in at least one image of a plurality of images;

determining, using the gesture recognition engine, a hand trajectory in the plurality of images;

determining, using the gesture recognition engine, a dynamic hand gesture in response to the hand posture and the hand trajectory;

providing, using the gesture recognition engine, a user interface command in response to determining the dynamic hand gesture;

wherein determining the hand trajectory comprises:

determining a plurality of hue-saturation-depth (HSD) histograms; and

tracking, using mean-shift analysis, a moving hand in response to the plurality of HSD histograms.

8. The non-transitory computer-readable storage medium of claim 7 , wherein determining the hand posture comprises:

detecting a hand in the at least one image;

segmenting, in response to detecting the hand, the at least one image into a binary image including a hand region; and

determining at least one shape feature in the hand region.

9. The non-transitory computer-readable storage medium of claim 8 , wherein the a least one shape feature comprises at least one of eccentricity, compactness, orientation, rectangularity, width center, height center, minimum box angle, minimum box width, number of defects, difference between left and right portions, or difference between top and bottom portions shape features.

10. The non-transitory computer-readable storage medium of claim 8 , wherein detecting the hand comprises using a cascade speeded-up robust feature (SURF) detector to detect the hand.

11. The non-transitory computer-readable storage medium of claim 7 , wherein determining the dynamic hand gesture comprises applying a hidden-markov model (HMM) to identify the dynamic hand gesture.

12. An apparatus, comprising:

a processor configured to:

detect an initiation gesture;

start a gesture recognition engine in response to detecting the initiation gesture;

determine, using the gesture recognition engine, a hand posture in at least one image of a plurality of images;

determine, using the gesture recognition engine, a hand trajectory in the plurality of images;

determine, using the gesture recognition engine, a dynamic hand gesture in response to the hand posture and the hand trajectory; and

provide, using the gesture recognition engine, a user interface command in response to determining the dynamic hand gesture;

wherein to determine the hand trajectory the processor is configured to:

determine a plurality of hue-saturation-depth (HSD) histograms; and

track, using mean-shift analysis, a moving hand in response to the plurality of HSD histograms.

13. The apparatus of claim 12 , wherein to determine the hand posture the processor is configured to:

detect a hand in the at least one image;

segment, in response to detecting the hand, the at least one image into a binary image including a hand region; and

determine at least one shape feature in the hand region.

14. The apparatus of claim 13 , wherein the a least one shape feature comprises at least one of eccentricity, compactness, orientation, rectangularity, width center, height center, minimum box angle, minimum box width, number of defects, difference between left and right portions, or difference between top and bottom portions shape features.

15. The apparatus of claim 13 , wherein to detect the hand the processor is configured to use a cascade speeded-up robust feature (SURF) detector.

16. The apparatus of claim 12 , wherein to determine the dynamic hand gesture the processor is configured to apply a hidden-markov model (HMM).

17. A system comprising:

an imaging device; and

a computing system, wherein the computing system is communicatively coupled to the imaging device and wherein, in response to depth images received from the imaging device, the computing system is to:

detect an initiation gesture;

start a gesture recognition engine in response to detecting the initiation gesture;

determine, using the gesture recognition engine, a hand posture in at least one image of a plurality of images;

determine, using the gesture recognition engine, a hand trajectory in the plurality of images;

determine, using the gesture recognition engine, a dynamic hand gesture in response to the hand posture and the hand trajectory;

provide, using the gesture recognition engine, a user interface command in response to determining the dynamic hand gesture;

wherein to determine the hand trajectory the computing system is to:

determine a plurality of hue-saturation-depth (HSD) histograms; and

track, using mean-shift analysis, a moving hand in response to the plurality of HSD histograms.

18. The system of claim 17 , wherein to determine the hand posture the computing system is to:

detect a hand in the at least one image;

segment, in response to detecting the hand, the at least one image into a binary image including a hand region; and

determine at least one shape feature in the hand region.

19. The system of claim 18 , wherein the a least one shape feature comprises at least one of eccentricity, compactness, orientation, rectangularity, width center, height center, minimum box angle, minimum box width, number of defects, difference between left and right portions, or difference between top and bottom portions shape features.

20. The system of claim 18 , wherein to detect the hand the computing system is to use a cascade speeded-up robust feature (SURF) detector.

21. The system of claim 17 , wherein to determine the dynamic hand gesture the computing system is to apply a hidden-markov model (HMM).